Willingness to pay for HIV self-tests among women in Kenya: implications for subsidy and pricing policies

Introduction

With roughly 30% of all people living with HIV globally remaining undiagnosed, innovative HIV testing approaches are essential for meeting the first of the UNAIDS “90-90-90” targets and achieving the benefits of treatment as prevention.¹ HIV self-testing (HIVST) is increasingly being recognized as an HIV testing approach that may appeal to hard-to-reach or high-risk individuals who would benefit from frequent HIV testing.² Based on research showing high acceptability and interest in HIVST across a wide range of populations and settings,^3–8 the WHO issued guidelines in December 2016 to support implementation and scale-up evidence-based HIVST approaches.⁹ Several countries in sub-Saharan Africa (SSA), including Kenya, have included HIVST in their testing guidelines and are currently developing plans to implement HIVST.^9,10 Governments and donors are now exploring HIVST distribution strategies – including retail distribution – that are best-suited to enhance HIV testing access among those not reached by existing HIV testing services (HTS).

Although self-test kits may become available in private sector pharmacies and other retail outlets, the extent to which they must be subsidized remains unknown. High prices in the retail sector may impede access in target populations, but very low prices or free distribution would require larger public subsidies and potentially mis-allocate resources by targeting those who already seek existing HTS. There have been few studies that have assessed individuals’ willingness to pay (WTP) for self-tests in SSA. Recent agreements between foundations and self-test manufacturers to achieve prices below current levels underscore the need for WTP data. To inform pricing policies for self-tests, we assessed WTP for self-tests among Kenyan women participating in a randomized trial.

Methods

Data were collected as part of a randomized trial to evaluate whether secondary distribution of self-tests can promote partner and couples testing (ClinicalTrials.gov NCT02386215).¹¹ Participants were 18–39 year-old women recruited between June 11, 2015 and January 16, 2016 from antenatal and postpartum clinics (ANC and PPC) in three urban and peri-urban health centers in Kisumu, Kenya. Participants were randomized to either an HIVST group that received two HIV self-tests free-of-charge or to a comparison group that received invitation cards for the male partner to come for clinic-based HIV testing. Self-tests given to participants in the HIVST group were oral fluid-based HIV test kits (OraQuick Rapid HIV-1/2 antibody tests, OraSure Technologies, Bethlehem, PA, USA).

At the time of enrollment, participants were administered a baseline questionnaire that collected information on socio-economic characteristics and health-related behaviors. Follow-up surveys were conducted with participants at 3 months to learn whether participants’ primary partner had an HIV test and whether the couple tested together. The follow-up survey also asked participants in the HIVST group whether they would be willing to pay for HIV self-tests, and if yes, what amount of money they would be willing to pay.

The primary outcome in this sub-study was a binary indicator of whether participants were willing to pay a non-zero amount for self-tests (i.e. WTP>0). We made the conservative assumption that participants who reported they were “not sure if they would pay for a self-test” had WTP=0. The secondary outcome was the amount, in Kenyan Shillings (KSH), that participants were willing to pay for self-tests. We assumed that participants who were willing to pay something but reported either “don’t know” for the WTP amount or refused to answer had a WTP equal to the median value among those with WTP>0. Various baseline characteristics of participants were used in our analyses. Categorical variables measured at baseline included educational attainment, occupation, self-reported chance of acquiring HIV in the future, and partner’s HIV testing history in the past year. Participants’ age, monthly income, and number of times tested for HIV in the past year were measured as continuous variables. Participants’ marital status, condom use at last sex, and intimate partner violence (IPV) history in the year prior were classified as binary variables.

Statistical analyses

We used a modified Poisson regression model to identify predictors of whether participants had WTP>0 for self-tests¹² and an ordinary least squares linear regression to identify predictors of the amount that participants were willing to pay for HIV self-tests. All models included robust standard errors as well as study site and interviewer fixed effects. Analyses were conducted in Stata 14.1.

The study received approval from the Scientific and Ethics Review Unit at the Kenya Medical Research Institute and the Office of Human Research Ethics at the University of North Carolina at Chapel Hill.

Results

In total, 297 participants were randomized to the HIVST group and 284 (96%) successfully completed a follow-up interview and provided responses to WTP questions. Four participants (1%) stated they were “not sure if they would pay for a self-test” and were assigned WTP=0. In addition, 23 participants (8%) were willing to pay something but reported either “don’t know” for the WTP amount or refused to answer altogether had a WTP equal to the median value among those with WTP>0. A vast majority (250/284, 88%) reported they would be willing to pay some amount of money for a self-test (Table 1). The median WTP amount among all participants was 100 KSH (IQR 50–150 KSH), or US$1 (IQR US$0.50–1.50). Participants with WTP=0 and WTP>0 had largely similar characteristics. Since participants were either pregnant or had recently delivered, most participants did not report any earnings in the past month. All participants who self-reported they had a high chance of acquiring HIV reported they were willing to pay for a self-test, whereas among those who reported a low chance or no chance, 16% (23/147) reported no WTP.

Table 1.

Participants’ willingness to pay for HIV self-tests

	WTP=0 KSH (n=34)	WTP>0 KSH (n=250)	Total (n=284)
Willingness to pay, 0 KSH	34 (100)		34 (12)
Willingness to pay, 1–50 KSH		66 (26)	66 (23)
Willingness to pay, 51–100 KSH		101 (40)	101 (36)
Willingness to pay, 100+ KSH		83 (33)	83 (29)
Willingness to pay, median (IQR)	0 (0–0)	100 (50–200)	100 (50–150)
Age, mean (SD)	23.4 (4.5)	24.3 (4.5)	0 (24.2)
Monthly earnings (KSH), median (IQR)	0 (0–1000)	0 (0–5000)	0 (0–4000)
Education:
Some primary or completed	20 (59)	123 (49)	143 (50)
Some secondary education	12 (35)	108 (43)	120 (42)
Completed secondary or greater	2 (6)	19 (8)	21 (7)
Lives with partner	33 (97)	236 (94)	269 (95)
Occupation:
Non-manual	6 (18)	77 (31)	83 (29)
Manual	3 (9)	25 (10)	28 (10)
Housewife/unemployed	25 (74)	148 (59)	173 (61)
Number of times been tested for HIV in last year, mean (SD)	2.8 (2)	2.8 (1.5)	2.8 (1.5)
Condom used during last sex	4 (12)	42 (17)	46 (16)
Experienced IPV in past year	8 (24)	70 (28)	78 (27)
Self-reported chance of HIV
High	0 (0)	9 (4)	9 (3)
Medium	3 (9)	30 (13)	33 (12)
Low	14 (44)	105 (44)	119 (44)
No risk at all	9 (28)	67 (28)	76 (28)
Don’t know	8 (24)	39 (16)	47 (17)
Partner tested in last year
Yes	4 (12)	38 (15)	42 (15)
No	14 (41)	135 (54)	149 (52)
Don’t know	16 (47)	77 (31)	93 (33)

Abbreviations: WTP, willingness to pay; SD, standard deviation; IQR, interquartile range; KSH, Kenya Shillings

For all variables frequencies are presented with percentages in parentheses except where otherwise noted

In regression analyses, the only variable significantly associated with WTP was participants’ self-reported chance of acquiring HIV at baseline. Those who reported a high chance were more likely to have WTP>0 than those who reported a low chance (risk ratio 1.21; 95% CI 1.02–1.44). There were no differences in WTP amounts, however, between participants with high and low self-reported chances of acquiring HIV. Participants who had tested more often in the past had significantly lower WTP, with each additional HIV test in the past 12 months being associated with a decrease of 16.3 KSH ($0.16) in WTP amount (P=0.001).

Discussion

Women who sought antenatal and postpartum services in the study and had gained experience using oral fluid-based self-tests reported being willing to pay some money for self-tests in the future. This suggests that self-test kits need not be fully subsidized in order to achieve reasonable levels of uptake. At the same time, participants’ WTP amounts indicated that demand will decline substantially if prices exceed US$1. Strengths of this study include its ascertainment of WTP from individuals who had a chance to become familiar with the new technologies and its focus on oral fluid-based test kits that are likely to become available for retail distribution. These findings could be relevant as the market for HIV self-tests takes shape, with donor agencies and governments making decisions on how much to subsidize the prices for these diagnostic products.

This study also helps predict the distributional implications of alternative HIVST pricing policies. Participants’ income and other socio-economic characteristics were not strong predictors of WTP, but rather participants’ self-assessed risk of acquiring HIV proved to be more relevant. Importantly, those who perceived themselves to be at high risk were more likely to be willing to pay some amount than those at low risk. Those who tested frequently in the past also had lower WTP. From an access standpoint, the findings from this study suggest that charging some money for self-tests may not exclude women at high risk or women with low utilization of existing HTS. The higher demand for self-tests in these groups may be due to both the financial barriers associated with clinic-based services as well as the greater convenience and privacy of self-tests.

While self-tests are not currently available for retail distribution in many countries, prices for oral fluid based self-tests in most low- and middle-income countries have exceeded US$7.50,¹³ an amount that few participants in our study were willing to pay. The recent decision by the Bill and Melinda Gates Foundation to subsidize OraQuick test kits and lower their price to US$2 in priority countries in SSA is an important advance in this regard.¹⁴ Based on our findings, however, a price of US$2 may be above the willingness to pay for many individuals. Additional subsidy from governments or greater economies of scale in production may be necessary to ensure high uptake.

This study has several limitations. First, WTP was assessed among women who received free self-tests and gained experience using them. Since familiarity with HIVST is low in much of SSA, WTP in the general population may be lower than it was in women who participated in this study and thus became familiar with self-tests. Supportive evidence stems from a recent study in coastal Kenya in which uptake of the oral self-tests was low among some pharmacy clients who were offered self-tests for US$1.¹⁵ Based on our study, the WTP amount of US$1 could be viewed as an upper bound of prices that should be charged in urban and peri-urban areas of Kenya where the study took place. Relatedly, it is likely that WTP will be lower in lower-income, rural populations. However, these findings may not generalize to other countries or even other parts of Kenya, where similar assessments of WTP will be necessary to gauge demand. Finally, WTP was self-reported by participants and thus may not be the same as their true WTP. The associations we found are nonetheless useful for identifying segments of the population that have higher demand for self-tests.

The pricing of diagnostic and preventive health products in poor countries is a topic that has received considerable attention in recent years. Several studies conducted in countries like Kenya and Zambia have used field-based pricing experiments in which prices are randomized to individuals in order to rapidly determine demand for preventive inputs like antimalarial bednets and water filters.^16,17 These studies have generally found a decline in demand with higher prices, suggesting that free distribution may be optimal. In some cases, however, small user fees have been rationalized as a way to increase revenue and reduce unnecessary use. While these products are very different from diagnostics such as HIV self-tests, methods used to estimate demand for these products can be used further explore the implications of alternative pricing strategies for self-tests. Additional studies exploring demand and WTP in different populations and settings are essential as countries seek to determine optimal subsidy levels for self-tests.